Les startups soutenues par des capital-risque commettent plus de fraudes, et les chercheurs pensent savoir pourquoi

VC-backed startups have been found to commit more fraud than non-venture-funded companies, according to a new report from the U.K.’s Imperial College and France’s Emlyon Business School. Published online in June, the research analyzed data from civil and criminal securities fraud prosecutions by the U.S. Securities and Exchange Commission (SEC) and Department of Justice (DOJ) between 2000 and 2023. The findings suggest that venture capital (VC)-backed founders are more prone to fraudulent behavior, particularly in environments where high-growth expectations are unmet. Among the notable cases highlighted in the report are convictions involving prominent figures such as Charlie Javice of FTX, Gökçe Güven of Kalder, Do Kwon of Terraform Labs, and Alexander and Valerie Lau Beckman of GameOn. These instances illustrate the scale and visibility of fraud within the tech sector. On platforms like X, the dominant social network for tech professionals, discussions around fraud and related terms like “scam” are frequent. Tim Weiss, one of the report's authors, noted that the normalization of fraud in the startup ecosystem is often overlooked, despite its prevalence. Another key study from the University of Toronto, also released in June, examined 654 fraud cases against U.S.-based VC-backed startups over the same period. While fraud is relatively uncommon overall, the research revealed that ventures receiving venture capital were significantly more likely to face fraud allegations. Startups founded during periods of excessive market enthusiasm, characterized by weak regulatory oversight and insufficient investor due diligence, were found to be 19% more likely to engage in fraudulent activities. Weiss explained that the issue extends beyond individual founders, pointing to the role of investors in shaping unrealistic expectations. He emphasized that the current surge in AI-driven startups mirrors historical patterns that have historically led to increased fraud. In these scenarios, the pressure to meet investor benchmarks can drive founders toward deceptive practices. In his paper, co-authored with Emlyon researcher Nevena Radoynovska, Weiss outlines a progression of fraudulent behaviors known as “façading.” This concept describes how founders might begin with subtle deceptions and gradually escalate to more elaborate falsehoods. Surface façading occurs when founders misrepresent their company’s progress or potential, especially during initial pitches to investors. This form of dishonesty goes beyond mere optimism or inflated market estimates. Reinforced façading involves the creation of fabricated evidence to substantiate earlier claims. An example cited in the paper includes a mobile testing app that generated fake customer contracts, invoices, and revenue figures to secure substantial investment at a unicorn valuation. As this pattern continues, founders may transition into deep façading, where they fabricate technological capabilities through staged demonstrations and maintain entirely false narratives about their operations. Despite these risks, the researchers observed that investors are not always passive participants. Some inadvertently contribute to fraudulent activity by setting excessively high performance targets. Weiss noted that while investors may unintentionally enable fraud, they also benefit from the outcomes presented by founders, who tailor results to align with investor expectations. The UT study further indicated that fraud accusations rarely hinder a founder’s ability to raise new capital. Even when fraud cases gain widespread media coverage, the VC community tends to overlook past misconduct. This resilience reflects a broader cultural tolerance for failure in Silicon Valley, where innovation is often celebrated regardless of the methods employed. Startups with founder-controlled boards were found to be twice as likely to engage in fraudulent activities compared to those with investor-controlled or shared control structures. This highlights the influence of governance models on ethical conduct within the startup ecosystem.